adding an interface to fill different types of containers from the compressed cluster...
[u/mrichter/AliRoot.git] / ANALYSIS / AliAnalysisAlien.h
c57f56b7 1#ifndef ALIANALYSISALIEN_H
3/* Copyright(c) 1998-1999, ALICE Experiment at CERN, All rights reserved. *
4 * See cxx source for full Copyright notice */
6// Author: Mihaela Gheata, 01/09/2008
9// AliAnalysisAlien - AliEn utility class. Provides interface for creating
10// a personalized JDL, finding and creating a dataset.
14#include "AliAnalysisGrid.h"
17#ifndef ROOT_TString
18#include <TString.h>
149d288c 21#ifndef ROOT_TMath
22#include <TMath.h>
28ce6c9c 25#ifndef ROOT_TMap
26#include <TMap.h>
105631b5 29class AliAnalysisManager;
77f34eae 30class AliAnalysisTaskCfg;
c57f56b7 31class TGridJDL;
33class AliAnalysisAlien : public AliAnalysisGrid {
37 AliAnalysisAlien();
38 AliAnalysisAlien(const char *name);
39 virtual ~AliAnalysisAlien();
40 AliAnalysisAlien(const AliAnalysisAlien& other);
41 AliAnalysisAlien& operator=(const AliAnalysisAlien& other);
42// Setters
f965131e 43 virtual void AddIncludePath(const char *path);
c57f56b7 44 virtual void AddRunNumber(Int_t run);
ee75cfc3 45 virtual void AddRunNumber(const char *run);
98690895 46 virtual void AddRunList(const char *runList);
c57f56b7 47 virtual void AddDataFile(const char *lfn);
648174cf 48 virtual void AddExternalPackage(const char *name);
c57f56b7 49 virtual void SetExecutable(const char *name="") {fExecutable = name;}
0a1c1f7f 50 virtual void SetExecutableCommand(const char *command="root -b -q") {fExecutableCommand = command;}
c57f56b7 51 virtual void SetArguments(const char *name="") {fArguments = name;}
631c0b05 52 virtual void SetExecutableArgs(const char *name="") {fExecutableArgs = name;}
c57f56b7 53 virtual void SetAnalysisMacro(const char *name="myAnalysis.C") {fAnalysisMacro = name;}
54 virtual void SetAnalysisSource(const char *name="myAnalysisClass.cxx") {fAnalysisSource = name;}
d3b18c4c 55 virtual void SetValidationScript(const char *name="") {fValidationScript = name;}
c57f56b7 56 virtual void SetAdditionalLibs(const char *list) {fAdditionalLibs = list;}
d5c6455a 57 virtual void SetAdditionalRootLibs(const char *list) {fAdditionalRootLibs = list;}
c57f56b7 58 virtual void SetPrice(Int_t price=1) {fPrice = price;}
a8739e8a 59 virtual void SetRunRange(Int_t min, Int_t max) {fRunRange[0] = min; fRunRange[1] = max;}
c6cb3634 60 virtual void SetJobTag(const char *tag="") {fJobTag = tag;}
319593fb 61 virtual void SetNrunsPerMaster(Int_t nruns=1) {fNrunsPerMaster = nruns;}
16a4353c 62 virtual void SetMaxMergeFiles(Int_t nfiles) {fMaxMergeFiles = nfiles;}
b3e07543 63 virtual void SetMaxMergeStages(Int_t nstages) {fMaxMergeStages = nstages;}
c57f56b7 64 virtual void SetSplitMode(const char *type="se") {fSplitMode = type;}
65 virtual void SetSplitMaxInputFileNumber(Int_t nfiles=100) {fSplitMaxInputFileNumber = nfiles;}
149d288c 66 virtual void SetAPIVersion(const char *version) {fAPIVersion = version;}
67 virtual void SetROOTVersion(const char *version) {fROOTVersion = version;}
68 virtual void SetAliROOTVersion(const char *version) {fAliROOTVersion=version;}
c57f56b7 69 virtual void SetUser(const char *user) {fUser = user;}
70 virtual void SetTTL(Int_t ttl=30000) {fTTL = ttl;}
71 virtual void SetGridWorkingDir(const char *name="workdir") {fGridWorkingDir = name;}
72 virtual void SetGridDataDir(const char *name) {fGridDataDir = name;}
73 virtual void SetDataPattern(const char *pattern="*AliESDs.root") {fDataPattern = pattern;}
0df6ccf2 74 virtual void SetFriendChainName(const char *name="") {fFriendChainName = name;}
bb885a9e 75 virtual void SetDefaultOutputs(Bool_t flag);
c57f56b7 76 virtual void SetGridOutputDir(const char *name="output") {fGridOutputDir = name;}
149d288c 77 virtual void SetOutputArchive(const char *list="*@disk=1*.root@disk=2");
78 virtual void SetOutputFiles(const char *list);
cd11251e 79 virtual void SetOutputToRunNo(Int_t mode=1) {fOutputToRunNo = mode;}
c57f56b7 80 virtual void SetInputFormat(const char *format="xml-single") {fInputFormat = format;}
81 virtual void SetMaxInitFailed(Int_t nfail=5) {fMaxInitFailed = nfail;}
d3b18c4c 82 virtual void SetTerminateFiles(const char *list) {fTerminateFiles = list;}
bb885a9e 83 virtual void SetMergeExcludes(const char *list) {fMergeExcludes = list;};
0f389141 84 virtual void SetMergeViaJDL(Bool_t on=kTRUE) {fMergeViaJDL = on ? 1 : 0;}
f47d5cb4 85 virtual void SetMergeDirName(const char *name) {fMergeDirName = name;}
c57f56b7 86 virtual void SetMasterResubmitThreshold(Int_t percentage) {fMasterResubmitThreshold = percentage;}
bb885a9e 87 virtual void SetNtestFiles(Int_t nfiles) {fNtestFiles = nfiles;}
149d288c 88 virtual void SetNumberOfReplicas(Int_t ncopies) {fNreplicas = TMath::Min(ncopies,4);}
c57f56b7 89 virtual void SetJDLName(const char *name="analysis.jdl") {fJDLName = name;}
149d288c 90 virtual void SetPreferedSE(const char *se);
a3e84053 91 virtual void SetProductionMode(Int_t mode=1) {fProductionMode = mode;}
74013535 92 virtual void SetRunPrefix(const char *prefix);
648174cf 93 virtual void SetOutputSingleFolder(const char *folder) {fOutputSingle = folder; fSplitMode="file"; fSplitMaxInputFileNumber=1;}
0f389141 94 virtual void SetFastReadOption(Bool_t on=kTRUE) {fFastReadOption = on ? 1 : 0;}
95 virtual void SetOverwriteMode(Bool_t on=kTRUE) {fOverwriteMode = on ? 1 : 0;}
c57f56b7 96
0f389141 97 TGridJDL *GetGridJDL() const {return fGridJDL;}
98 TGridJDL *GetMergingJDL() const {return fMergingJDL;}
5513444a 99 const char *GetGridOutputDir() const {return fGridOutputDir;}
c57f56b7 100//Utilities
77f34eae 101 void AddModule(AliAnalysisTaskCfg *module);
102 void AddModules(TObjArray *list);
3e40fd4c 103 AliAnalysisManager *CreateAnalysisManager(const char *name, const char *filename="");
77f34eae 104 Int_t GetNmodules() const;
105 AliAnalysisTaskCfg *GetModule(const char *name);
106 Bool_t LoadModules();
21ca8e59 107 Bool_t GenerateTest(const char *name, const char *modname="");
348be253 108 Bool_t CheckFileCopy(const char *alienpath);
c57f56b7 109 virtual Bool_t CreateDataset(const char *pattern);
21ca8e59 110 Bool_t CopyLocalDataset(const char *griddir, const char *pattern, Int_t nfiles, const char *output="data.txt", const char *anchorfile="");
c57f56b7 111 virtual Bool_t CreateJDL();
4e5c5506 112 virtual void EnablePackage(const char *package);
923e2ca5 113 static Bool_t DirectoryExists(const char *lfn);
5513444a 114 static Bool_t FileExists(const char *lfn);
a03be957 115 static const char *GetJobStatus(Int_t jobidstart, Int_t lastid, Int_t &nrunning, Int_t &nwaiting, Int_t &nerror, Int_t &ndone);
f790bc1b 116 const char *GetListOfFiles(const char *type);
b3e07543 117 Bool_t CheckMergedFiles(const char *filename, const char *aliendir, Int_t nperchunk, const char *jdl="");
105631b5 118 static AliAnalysisManager *LoadAnalysisManager(const char *fname);
b3e07543 119 static Bool_t MergeOutput(const char *output, const char *basedir, Int_t nmaxmerge, Int_t stage=0);
c57f56b7 120 virtual Bool_t MergeOutputs();
16a4353c 121 virtual void Print(Option_t *option="") const;
5513444a 122 virtual Bool_t StartAnalysis(Long64_t nentries=123456789, Long64_t firstentry=0);
0f389141 123 static Bool_t SetupPar(const char *package);
a03be957 124 virtual Bool_t Submit();
125 virtual Bool_t SubmitMerging();
7c2cd90a 126 static Int_t SubmitSingleJob(const char *query);
c57f56b7 127 virtual void WriteAnalysisFile();
128 virtual void WriteAnalysisMacro();
0f389141 129 virtual void WriteMergingMacro();
130 virtual void WriteMergeExecutable();
c57f56b7 131 virtual void WriteExecutable();
5513444a 132 virtual Bool_t WriteJDL(Bool_t copy);
133 virtual void WriteProductionFile(const char *filename) const;
0f389141 134 virtual void WriteValidationScript(Bool_t merge=kFALSE);
c57f56b7 135
3bdcb562 136// PROOF mode
137 virtual void SetProofCluster(const char *cluster) {fProofCluster = cluster;}
138 virtual void SetProofDataSet(const char *dataset) {fProofDataSet = dataset;}
139 virtual const char *GetProofDataSet() const {return fProofDataSet.Data();}
28ce6c9c 140 virtual void SetProofParameter(const char *pname, const char *value);
141 const char *GetProofParameter(const char *pname) const;
3bdcb562 142 virtual void SetProofReset(Int_t mode) {fProofReset = mode;}
143 virtual void SetNproofWorkers(Int_t nworkers) {fNproofWorkers = nworkers;}
27734f0e 144 virtual void SetNproofWorkersPerSlave(Int_t nworkers) {fNproofWorkersPerSlave = nworkers;}
3bdcb562 145 virtual void SetRootVersionForProof(const char *version) {fRootVersionForProof = version;}
146 virtual void SetAliRootMode(const char *mode) {fAliRootMode = mode;}
147 // .txt file containing the list of files to be chained in test mode
148 virtual void SetFileForTestMode(const char *filename) {fFileForTestMode = filename;}
149 virtual TChain *GetChainForTestMode(const char *treeName) const;
c57f56b7 151protected:
152 void CdWork();
153 Bool_t CheckInputData();
154 void CheckDataType(const char *lfn, Bool_t &is_collection, Bool_t &is_xml, Bool_t &use_tags);
155 virtual Bool_t Connect();
156 virtual void SetDefaults();
a03be957 157 Bool_t SubmitNext();
c57f56b7 158
c57f56b7 159 Bool_t IsCollection(const char *lfn) const;
fe2d7fc2 160 virtual Bool_t IsSingleOutput() const;
c57f56b7 161 Bool_t IsUsingTags() const {return TObject::TestBit(AliAnalysisGrid::kUseTags);}
77f34eae 162 Bool_t LoadModule(AliAnalysisTaskCfg *mod);
163 Bool_t CheckDependencies();
c57f56b7 164
166 TGridJDL *fGridJDL; //! JDL maker
0f389141 167 TGridJDL *fMergingJDL; //! JDL maker
c57f56b7 168 Int_t fPrice; // Grid price for the job;
169 Int_t fTTL; // Time to live.
170 Int_t fSplitMaxInputFileNumber; // Maximum number of files to be processed per subjob
171 Int_t fMaxInitFailed; // Maximum initial consecutive subjobs accepted to fail
172 Int_t fMasterResubmitThreshold; // Failed jobs will be resubmitted until this DONE ratio
bb885a9e 173 Int_t fNtestFiles; // Number of files used in the testing case
a8739e8a 174 Int_t fRunRange[2]; // Run range
319593fb 175 Int_t fNrunsPerMaster; // Number of runs per masterjob
16a4353c 176 Int_t fMaxMergeFiles; // Maximum number of files to be merged in one chunk
b3e07543 177 Int_t fMaxMergeStages; // Maximum number of merging stages
d2a409b2 178 Int_t fNsubmitted; // Number of jobs submitted
a3e84053 179 Int_t fProductionMode; // Production mode (0-off, 1-on)
cd11251e 180 Int_t fOutputToRunNo; // Use run number as output directory
0f389141 181 Int_t fMergeViaJDL; // Enable merging via automatic JDL
182 Int_t fFastReadOption; // Use xrootd tweaks to reduce timeouts in file access
183 Int_t fOverwriteMode; // Overwrite existing files if any
149d288c 184 Int_t fNreplicas; // Number of replicas for the output files
3bdcb562 185 Int_t fNproofWorkers; // Number of workers in proof mode
27734f0e 186 Int_t fNproofWorkersPerSlave; // Max number of workers per slave in proof mode
3bdcb562 187 Int_t fProofReset; // Proof reset mode: 0=no reset, 1=soft, 2=hard
c57f56b7 188 TString fRunNumbers; // List of runs to be processed
189 TString fExecutable; // Executable script for AliEn job
0a1c1f7f 190 TString fExecutableCommand; // Command(s) to be executed in the executable script
c57f56b7 191 TString fArguments; // Arguments for the executable script
631c0b05 192 TString fExecutableArgs; // arguments added to the executable script after the analysis macro
c57f56b7 193 TString fAnalysisMacro; // Root macro steering the analysis
194 TString fAnalysisSource; // User analysis implementation (.cxx) file(s)
d3b18c4c 195 TString fValidationScript; // Name of the validation script
d5c6455a 196 TString fAdditionalRootLibs; // List (separated by blacs) of additional libraries needed for/before analysis libs/par file compilation
197 TString fAdditionalLibs; // List (separated by blacs) of additional libraries needed for the analysis loaded AFTER all par files
c57f56b7 198 TString fSplitMode; // Job split mode
199 TString fAPIVersion; // API version
200 TString fROOTVersion; // ROOT version
201 TString fAliROOTVersion; // AliROOT version
648174cf 202 TString fExternalPackages; // External packages
c57f56b7 203 TString fUser; // AliEn user name
204 TString fGridWorkingDir; // AliEn directory containing the input packages
205 TString fGridDataDir; // AliEn data production directory
206 TString fDataPattern; // Data pattern for 'find' command
207 TString fGridOutputDir; // AliEn directory (wrt work dir) where the output should be written
208 TString fOutputArchive; // List of output archives separated by blancs
209 TString fOutputFiles; // List of output files separated by blancs
210 TString fInputFormat; // Input format (xml-single)
211 TString fDatasetName; // Dataset xml file to be created
212 TString fJDLName; // JDL file to be generated
d3b18c4c 213 TString fTerminateFiles; // List of output files produced during Terminate
bb885a9e 214 TString fMergeExcludes; // List of output files excluded from merging
f965131e 215 TString fIncludePath; // Include path
bb885a9e 216 TString fCloseSE; // Preffered storage element. Taken from alien_CLOSE_SE environment.
0df6ccf2 217 TString fFriendChainName; // File name to construct friend chain (for AOD)
c6cb3634 218 TString fJobTag; // Job tag
648174cf 219 TString fOutputSingle; // Directory name for the output when split is per file
5fce53f4 220 TString fRunPrefix; // Run prefix to be applied to run numbers
3bdcb562 221 TString fProofCluster; // Proof cluster name
222 TString fProofDataSet; // Proof dataset to be used
223 TString fFileForTestMode; // .txt file for the chain to be used in PROOF test mode
224 TString fRootVersionForProof; // ROOT version to be used in PROOF mode. The default one taken if empty.
225 TString fAliRootMode; // AliRoot mode among the list supported by the proof cluster
f47d5cb4 226 TString fMergeDirName; // Name of the directory that should be added to the output directory
c57f56b7 227 TObjArray *fInputFiles; // List of input files to be processed by the job
4e5c5506 228 TObjArray *fPackages; // List of packages to be used
77f34eae 229 TObjArray *fModules; // List of AliAnalysisTaskCfg modules
28ce6c9c 230 TMap fProofParam; // Key-value pairs for proof mode
c57f56b7 231
77f34eae 232 ClassDef(AliAnalysisAlien, 19) // Class providing some AliEn utilities
c57f56b7 233};